How a fish tells its own kind from a stranger, and why a new book is rebuilding the molecular-biology bench from scratch

Fish can tell each other apart. That sounds obvious, until you try to explain the mechanism.

In a feature published 15 July 2026, New Scientist lays out a problem that has quietly occupied evolutionary biologists for decades: how a single fish, swimming through turbid water crowded with look-alike neighbours, decides which other fish is one of its own. The question turns out to be a tangle of visual perception, olfactory signalling, hormonal state, social learning and, increasingly, the kind of multisensory integration that researchers only learned to measure in the last ten years.

The implications stretch well beyond ichthyology. Species recognition is the gatekeeper of mate choice, of hybridisation, of inbreeding avoidance. Get it wrong, and a lineage can collapse or, conversely, fuse with a neighbour. Understanding the machinery is, in practice, understanding the engine of speciation itself.

The sensory problem, in plain terms

Researchers interviewed by New Scientist describe a layered recognition system. Vision does the first sweep, often relying on species-specific colour patterns, body stripes or fin shapes. Cichlids in the African Great Lakes are the textbook case: hundreds of closely related species packed into the same water, distinguishable mostly by the colours on their flanks. Olfaction does the fine work, picking up chemical cues that visual systems cannot resolve at distance or in low light. A third layer, behavioural, takes over once the fish is close enough to read posture and movement.

None of these layers operate in isolation. A cichlid that has been chemically exposed to another species' pheromones will, in lab trials, alter its visual preferences. Hormonal state changes the salience of certain cues. Early social environment, who a juvenile fish grew up around, recalibrates the whole apparatus. The result is a recognition system that is less a checklist and more a weighted consensus across senses, life stage and experience.

This is why the field has resisted clean answers. A trait that looks decisive in the lab, say, a particular stripe pattern, may matter far less in a turbid river where chemical cues dominate. The same species can recognise each other by entirely different routes in different habitats.

Why the textbook answer keeps slipping

For most of the twentieth century, the dominant explanation was simple: a few genes of large effect, often grouped under the loose banner of "magic traits," drove both mate choice and ecological divergence at the same time. A fish that prefers, and is preferred by, its own kind, the story went, would be naturally selected to look, smell and behave distinctively. Speciation followed almost as a side effect.

The newer picture is messier. Genome-wide studies have found that many of the signals fish use to identify each other sit in regions of the genome shaped by selection, but the selection is often weak, polygenic, and environmentally contingent. The clean magic-trait model still has defenders, but it is no longer the default. New Scientist's feature sits inside that shift: away from a single-gene story, toward a sensory-integration story in which the genome supplies the palette and the environment, the developmental stage and the social context pick the colours.

The practical upshot is that conservation biologists working on hybridising populations, salmon in Scotland, cichlids in Lake Victoria, pupfish in the American Southwest, can no longer assume that one diagnostic marker will tell them who is breeding with whom. They need to measure behaviour and signal in the field, not just genotype in the lab.

A book that takes the lab itself as the subject

On the publishing side, a different kind of history is being written, from the bench outward. On 15 July 2026, Niko McCarty, writing on X, announced that an Asimov Press book on the history of the molecular-biology laboratory was nearly finished. The manuscript, McCarty wrote, sits at 455 pages and includes more than 100 images, accumulated over six-plus months of writing and editing.

The subject is the room itself: the bench, the centrifuge, the agar plate, the pipette tip box, the autoclave, the shared reagent stock. McCarty's framing, drawn from the previews shared in the thread, treats the laboratory not as a backdrop to discovery but as a technology in its own right, one that co-evolved with the questions biologists were able to ask. The polymerase chain reaction, the Sanger sequencer, the microtitre plate, the cryo-electron microscope: each of these is both an instrument and an argument about what kind of biology is worth doing.

That framing matters because the molecular-biology lab of 2026 looks almost nothing like the molecular-biology lab of 1966, and the change is not only a matter of better machines. The division of labour has shifted. A 1970s lab was a workshop: a principal investigator, a bench or two of postdocs and graduate students, glassware washed by hand, reagents mixed in-house. A 2020s lab is closer to a small factory floor: automated liquid handlers, commercial kits, shared core facilities, datasets that travel between continents. The book's argument, as previewed, is that the institutional shape of the lab, who works in it, who pays for it, and what counts as a publishable result, has driven the science as much as any individual genius.

Stakes and what to watch next

Two storylines, one methodological, one historiographical, are running in parallel. The fish-recognition work is forcing biologists to expand their unit of analysis from the gene to the integrated sensory system, and from the controlled lab to the variable field. The Asimov Press project is doing something similar from the other end: forcing historians of science to expand their unit of analysis from the scientist to the room the scientist works in.

What remains uncertain in both cases is the level of resolution that will hold up. The New Scientist feature flags that several proposed mechanisms for species recognition, particularly the role of early social learning, are still based on small sample sizes and a handful of model species. Generalising beyond cichlids and sticklebacks is, for now, an act of faith. On the book side, McCarty has announced a near-finished manuscript, not a published one: the editorial and peer review process, and what survives of it, will shape which version of the lab's history actually reaches readers.

For now, both stories point in the same direction. The clean, single-cause explanations, the one-gene trait, the lone-genius discovery, are giving way to descriptions that take the whole system seriously: the fish in its water, the scientist at the bench, the institution that pays for the centrifuge.

Desk note: this piece pairs two science desk threads that arrived within hours of each other on 15 July 2026. Monexus ran them as a single article because both are about the same underlying shift, away from single-cause explanations and toward systems-level accounts of how biological order gets made and maintained.